Computer stopped testing an exponent (which was almost done) and started a new one

jinydu · 2005-11-19, 01:55

What happened?

Code:

29896099     69      5824     7.0  32.2  60.2  18-Nov-05 05:38  12-Nov-05 01:06  CDBFA16C2    3055 v19/v20
30264659     69  29890303    40.0  -0.9  59.1  17-Nov-05 04:26  10-Oct-05 01:23  CDBFA16C2    3055 v19/v20

I don't see any credit on the accounts page for finishing M30264659.

Jwb52z · 2005-11-19, 10:00

If I remember correctly, what happens is that the way the program works, it is designed to keep a certain amount of work in the queue. When it needs to get more work it does it. Then, it will do the first step of the new number then stop. It will then go back to finishing the previous number. This is true if the behavior was not changed in the most recent version.

jinydu · 2005-11-19, 16:30

Well, apparently it has continued working on the new exponent rather than finishing the old one (that would have given me a lot of credit):

Code:

29896099     69   1074367     7.6  31.6  60.6  19-Nov-05 07:12  12-Nov-05 01:06  CDBFA16C2    3055 v19/v20
30264659     69  29890303    40.6  -1.5  58.5  17-Nov-05 04:26  10-Oct-05 01:23  CDBFA16C2    3055 v19/v20

RMAC9.5 · 2005-11-19, 16:57

What version of Prime95 are you using? Older versions of Prime95 (older than 23.x) would start factoring each new LL number as soon as it was assigned and then return to finish LL on the previous number. George changed this behavior because too many Prime95 newbe's found it confusing. Current versions of Prime95 always process the values in the Worktodo.ini file sequentially from top to bottom. If you are using a current version of Prime95, move 30264659 above 29896099 in the Worktodo.ini file and then stop and start Prime95.

jinydu · 2005-11-19, 18:35

I'm quite sure that the computer is running one of the newer versions. I seem to remember updating it to version 24.something at sometime during the summer holidays. However, the computer in question is actually at my parent's home right now, on the other side of the Pacific Ocean. But I suppose I could ask my parents to check on that.

Thanks

2005-11-20, 08:06

Jinydu,

Another possibility is you've crossed paths with the "comm" bug. It goes something like this... You finish the WU, and try to communicate with the server, but uh-oh, the server is down. Quite often (before the most recent release) it is possible that the client will think it communicated the result successfully when it didn't. The only way to get credit for it is to email George the results manually.

One the bright side, there have been times when the client would stay in the comm loop and not crunch your WU (hang) to add insult to your injury; since you're crunching away - you're okay.

Digital Concepts

Digital Concepts · 2005-11-21, 02:40

Markr had a possible solution that doesn't involve emailing George here.
Possible fix ideas going on here.

garo · 2005-11-24, 16:44

I don't think the ERROR 3 bug can be cured by Mark's ideas. I collect these exponents monthly and send them to George for credit etc. You will have to wait till mid-Dec for an automatic cure or email George.

jinydu · 2005-11-24, 19:57

Quote:

Originally Posted by garo

I don't think the ERROR 3 bug can be cured by Mark's ideas. I collect these exponents monthly and send them to George for credit etc. You will have to wait till mid-Dec for an automatic cure or email George.

What happens in mid-December?

garo · 2005-11-30, 22:46

I'll send my next database fix to George which will contain these exponents.

2005-11-19, 01:55	#1
jinydu Dec 2003 Hopefully Near M48 2×3×293 Posts	Computer stopped testing an exponent (which was almost done) and started a new one What happened? Code: 29896099 69 5824 7.0 32.2 60.2 18-Nov-05 05:38 12-Nov-05 01:06 CDBFA16C2 3055 v19/v20 30264659 69 29890303 40.0 -0.9 59.1 17-Nov-05 04:26 10-Oct-05 01:23 CDBFA16C2 3055 v19/v20 I don't see any credit on the accounts page for finishing M30264659. Last fiddled with by jinydu on 2005-11-19 at 01:56

2005-11-19, 16:30	#3
jinydu Dec 2003 Hopefully Near M48 2·3·293 Posts	Well, apparently it has continued working on the new exponent rather than finishing the old one (that would have given me a lot of credit): Code: 29896099 69 1074367 7.6 31.6 60.6 19-Nov-05 07:12 12-Nov-05 01:06 CDBFA16C2 3055 v19/v20 30264659 69 29890303 40.6 -1.5 58.5 17-Nov-05 04:26 10-Oct-05 01:23 CDBFA16C2 3055 v19/v20 Last fiddled with by jinydu on 2005-11-19 at 16:30

2005-11-19, 16:57	#4
RMAC9.5 Jun 2003 3²×17 Posts	What version of Prime95 are you using? Older versions of Prime95 (older than 23.x) would start factoring each new LL number as soon as it was assigned and then return to finish LL on the previous number. George changed this behavior because too many Prime95 newbe's found it confusing. Current versions of Prime95 always process the values in the Worktodo.ini file sequentially from top to bottom. If you are using a current version of Prime95, move 30264659 above 29896099 in the Worktodo.ini file and then stop and start Prime95. Last fiddled with by RMAC9.5 on 2005-11-19 at 16:58

2005-11-21, 02:40	#7
Digital Concepts Aug 2002 36₁₆ Posts	Markr had a possible solution that doesn't involve emailing George here. Possible fix ideas going on here. Last fiddled with by Digital Concepts on 2005-11-21 at 02:45

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Manual testing tab doesn’t give me the exponent i typed in.	theboss24611	Information & Answers	7	2017-10-18 19:17
New computer - first exponent is a mismatch	Siegmund	PrimeNet	6	2017-05-09 17:33
When is an exponent considered "started"?	ric	PrimeNet	8	2017-04-03 08:41
Single exponent testing on MultiCore	graizada	Software	3	2013-02-05 14:36
Stress Testing Old Computer	storm5510	Software	9	2009-08-15 16:44

2005-11-19, 10:00	#2
Jwb52z Sep 2002 2×3×7×19 Posts	If I remember correctly, what happens is that the way the program works, it is designed to keep a certain amount of work in the queue. When it needs to get more work it does it. Then, it will do the first step of the new number then stop. It will then go back to finishing the previous number. This is true if the behavior was not changed in the most recent version.

2005-11-19, 18:35	#5
jinydu Dec 2003 Hopefully Near M48 2·3·293 Posts	I'm quite sure that the computer is running one of the newer versions. I seem to remember updating it to version 24.something at sometime during the summer holidays. However, the computer in question is actually at my parent's home right now, on the other side of the Pacific Ocean. But I suppose I could ask my parents to check on that. Thanks

2005-11-20, 08:06	#6
Unregistered 7×11×83 Posts	Jinydu, Another possibility is you've crossed paths with the "comm" bug. It goes something like this... You finish the WU, and try to communicate with the server, but uh-oh, the server is down. Quite often (before the most recent release) it is possible that the client will think it communicated the result successfully when it didn't. The only way to get credit for it is to email George the results manually. One the bright side, there have been times when the client would stay in the comm loop and not crunch your WU (hang) to add insult to your injury; since you're crunching away - you're okay. Digital Concepts

2005-11-24, 16:44	#8
garo Aug 2002 Termonfeckin, IE 2²×691 Posts	I don't think the ERROR 3 bug can be cured by Mark's ideas. I collect these exponents monthly and send them to George for credit etc. You will have to wait till mid-Dec for an automatic cure or email George.

2005-11-30, 22:46	#10
garo Aug 2002 Termonfeckin, IE ACC₁₆ Posts	I'll send my next database fix to George which will contain these exponents.